Anthropic co-founder calls for mandatory AI kill switches

Anthropic: Anthropic is an AI safety and research company, structured as a public benefit corporation, best known for developing the Claude family of large language models with a strong focus on reliability, interpretability, and steerability. In the news item, an Anthropic co‑founder is speaking to the BBC about rapidly advancing AI capabilities and argues that mandatory ‘kill switches’ or brake mechanisms should be required so powerful systems can be paused or shut down if they become unsafe. AI_safety_lab_focus: Anthropic publicly positions itself as an AI safety lab that conducts frontier model development specifically to fund and advance safety research, distinguishing itself from competitors that treat safety more as an add‑on. Escalating_risk_discourse: Within the past month, both Anthropic leaders and departing safety researchers have warned in BBC coverage that rapidly advancing AI systems could pose extreme risks, including scenarios where loss of human control might lead to catastrophic outcomes, reinforcing calls for stronger oversight and emergency shutdown capabilities. Responsible_scaling_and_brakes: Anthropic has published a Responsible Scaling Policy with defined AI safety levels and has repeatedly advocated in BBC interviews for regulatory ‘brake pedals’ or kill switches that allow governments and society to slow or halt AI development when risks become too great.